Data Reduction Techniques for Real-time Fault Detection and Diagnosis, and Multiple Fault Inference with Imperfect Tests

Limited downlink data rate constrains the amount of data that can be sent to earth from a spacecraft. Data from the onboard health monitoring sensors needs to be accommodated within a small fraction of this downlink bandwidth. The problem is more acute for interplanetary missions, where the downlink data rate is significantly lower than the low earth orbit missions. Such constraint prohibits transmission of the complete set of health monitoring data. This proposed Phase-II effort is geared towards providing enhanced remote diagnostics using limited telemetry bandwidth. Diagnostic accuracy of health management system depends more on the information content of the monitored data, than on its sheer volume. We propose accommodating more information within the allocated downlink bandwidth for health monitoring by performing intelligent data reduction. The onboard data reduction process employs sensor fusion, dimensionality reduction and temporal fusion techniques.